Indexing Tools

Prerequisites

Credential for your toolkit (required for all except Artifact). See Create a Credential.
Vector Storage: PgVector selected in Settings → AI Configuration.
Embedding Model selected in Settings → AI Configuration.
A configured toolkit that supports indexing. See Indexing Overview.

Index Data

Create a new index or refresh an existing one from the connected source system. This tool collects source content, processes it with the configured chunking strategy, and stores the result in the vector database so it can be searched later. Purpose

Create a searchable index from supported toolkit data.
Update an existing index with the latest source content.
Prepare indexed content for Search Index, Stepback Search Index, and Stepback Summary Index.

How to Execute The Index Data tool runs when you create an index. Recommended: Indexes Section

Open your toolkit.
Expand the Indexes section.
Select + Index.
Fill in the Index Data parameters.
Select Index to run the tool and create the index.

Common Settings & Parameters

UI label	Description	Required?	Default value	Validation/Allowed
Index Name	Unique collection name for the index	Yes	None	max_length=7, min_length=1
Progress Step	Step size for progress reporting	No	10	ge=0, le=100
Clean Index	Clean existing index before re-indexing	No	`FALSE`	None
Skip Unsupported Extensions	Skip files with unsupported extensions instead of processing them with the text chunker	No	`TRUE`	boolean
Chunking Config	Configuration for chunking tool	No	`{}`	None

New Index page showing the full index configuration form.

Toolkit-Specific Settings & Parameters

The following parameters apply to specific toolkit types only.

Toolkit type	UI label	Description	Required?	Default	Validation/Allowed
Repositories (GitHub, GitLab, Bitbucket, ADO Repos)	Branch	Branch to index files from. Defaults to the active branch if not provided.	No	None	string
Repositories (GitHub, GitLab, Bitbucket, ADO Repos)	Whitelist	File extensions or paths to include. Example: `[".md", ".java"]`.	No	None	array of glob patterns
Repositories (GitHub, GitLab, Bitbucket, ADO Repos)	Blacklist	File extensions or paths to exclude. Example: `[".md", ".java"]`.	No	None	array of glob patterns
Repositories (GitHub, GitLab, Bitbucket, ADO Repos)	Skip Unsupported Extensions	Skip files with unsupported extensions instead of processing them with the text chunker.	No	`TRUE`	boolean
Test Management (Zephyr Enterprise)	Zql	ZQL query to search for test cases. Supported fields include estimatedTime, testcaseId, creator, release, project, priority, altId, version, versionId, automated, folder, contents, name, comment, and tag. Examples: `folder="TestToolkit"`, `name~"TestToolkit5"`.	No	None	ZQL syntax
Test Management (Zephyr Scale)	Project Key	Jira project key filter (for example, `PROJ`).	Yes	None	Jira project key
Test Management (Zephyr Scale)	Jql	JQL-like query for searching test cases. Supported fields include folder, folderPath, label, text, customFields, steps, orderBy, orderDirection, limit, includeSubfolders, and exactFolderMatch.	Yes	None	JQL-like syntax
ADO Boards	Wiql	WIQL (Work Item Query Language) query string to select and filter Azure DevOps work items.	Yes	None	WIQL syntax
ADO Boards	Workers	Maximum number of work items processed concurrently.	No	1	integer, 1–10
ADO Boards	Process Images	Describe inline images in HTML work item fields with the LLM before indexing.	No	`FALSE`	boolean
ADO Boards	Image Description Prompt	Optional prompt used when `process_images` is turned on.	No	None	string
ADO Boards	Fields	Whitelist of work item field reference names to include in indexed content.	No	None	array of field names
ADO Boards	Sanitize	Sanitize HTML-rich field content before indexing.	No	`TRUE`	boolean
ADO Plans	Plan Id	ID of the test plan for which test cases are requested.	Yes	None	numeric ID
ADO Plans	Suite Ids	List of test suite IDs for which test cases are requested. Leave empty to index all suites from the plan. Example: `[2, 23]`.	No	`[]`	array of numeric IDs
ADO Wiki	Wiki Identifier	Wiki identifier to index (for example, `ABCProject.wiki`). If omitted, the default wiki identifier from the toolkit configuration is used.	No	None	string
ADO Wiki	Path Contains	Case-insensitive substring filter applied to the full wiki page path. Leave empty to index all pages.	No	None	string
ADO Wiki	Process Images	Replace inline image markdown with LLM-generated descriptions so image content becomes searchable text.	No	`FALSE`	boolean
ADO Wiki	Image Description Prompt	Optional custom prompt used when generating image descriptions.	No	None	string
ADO Wiki	Include Attachments	Index `/.attachments/` files referenced by wiki pages as dependent documents.	No	`FALSE`	boolean
SharePoint	Limit Files	Maximum number of files to return. Supports synonyms such as First, Top, or a numeric literal (for example, `Top 10 files`). If not specified, the default applies.	No	1000	number or synonym keyword
SharePoint	Include Extensions	List of file extensions to include when processing. If empty, all files are processed except those in Skip Extensions. Example: `[".png", ".jpg"]`.	No	None	array of glob patterns
SharePoint	Skip Extensions	List of file extensions to skip when processing. Example: `[".png", ".jpg"]`.	No	None	array of glob patterns
SharePoint	Path	Folder path to scope indexing. Accepts a full server-relative path or a relative path.	No	None	string
SharePoint	Form Name	Document library name to scope indexing to.	No	None	string
SharePoint	Include OneNote	Index OneNote pages from the SharePoint site. Requires Graph API delegated access.	No	`FALSE`	boolean
SharePoint	OneNote Filter	Optional dictionary controlling which OneNote notebooks, sections, and pages are indexed and how they are processed.	No	None	dictionary
Figma	Urls Or File Keys	Comma-separated list of Figma file URLs or raw file keys to index.	Yes	None	comma-separated string
Figma	Number Of Threads	Number of worker threads used when indexing Figma images.	No	5	integer, 1–5
Figma	Node Ids Include	List of top-level node IDs (pages) to include in the index.	No	None	array of strings
Figma	Node Ids Exclude	List of top-level node IDs (pages) to exclude when `node_ids_include` is not provided.	No	None	array of strings
Figma	Node Types Include	List of node types to include in the index.	No	None	array of strings
Figma	Node Types Exclude	List of node types to exclude when `node_types_include` is not provided.	No	None	array of strings
Figma	Scale To Limit Threshold	Maximum image dimension in pixels for LLM vision processing. Larger frames are scaled down.	No	8000	integer
Figma	Subframe Extract Threshold	Frames larger than this dimension are split into subframes and processed individually.	No	15000	integer
TestRail	Project Id	TestRail project ID to index data from.	Yes	None	string or numeric ID
TestRail	Suite Id	Optional TestRail suite ID to filter test cases.	No	None	string or numeric ID
TestRail	Section Id	Optional section ID to filter test cases.	No	None	numeric ID
TestRail	Include Attachments	Whether to include attachment content in indexing.	No	`FALSE`	boolean
TestRail	Skip Attachment Extensions	List of file extensions to skip when processing attachments (for example, `['.png', '.jpg']`).	No	None	array of extensions
Xray Cloud	Jql	JQL query for searching test cases in Xray. Supported fields include project, testType, labels, summary, description, status, and priority.	No	None	JQL syntax
Xray Cloud	Graphql	Custom GraphQL query for advanced data extraction. Use either `jql` or `graphql`, not both.	No	None	GraphQL syntax
Xray Cloud	Include Attachments	Whether to include attachment content in indexing.	No	`FALSE`	boolean
Xray Cloud	Skip Attachment Extensions	List of file extensions to skip when processing attachments (for example, `['.exe', '.zip', '.bin']`).	No	None	array of extensions
Confluence	Content Format	Render format for page content.	No	`view`	string (for example, `view`)
Confluence	Page Ids	List of page IDs to retrieve.	No	None	array of IDs
Confluence	Label	Label to filter pages.	No	None	string
Confluence	Cql	CQL query to filter pages.	No	None	CQL syntax
Confluence	Limit	Limit the number of results.	No	10	number
Confluence	Max Pages	Maximum number of pages to retrieve.	No	1000	number
Confluence	Include Restricted Content	Include restricted content.	No	`FALSE`	boolean
Confluence	Include Archived Content	Include archived content.	No	`FALSE`	boolean
Confluence	Include Attachments	Whether to include attachment content in indexing.	No	`FALSE`	boolean
Confluence	Include Extensions	List of file extensions to include when processing attachments. If empty, all files are processed except those in Skip Extensions.	No	`[]`	array of glob patterns
Confluence	Skip Extensions	List of file extensions to skip when processing attachments.	No	`[]`	array of glob patterns
Confluence	Include Comments	Include page comments in indexing.	No	`FALSE`	boolean
Confluence	Include Labels	Include page labels in indexing.	No	`FALSE`	boolean
Confluence	Ocr Languages	OCR languages for processing attachments.	No	`eng`	language codes (for example, `eng`)
Confluence	Keep Markdown Format	Keep Markdown formatting in output.	No	`TRUE`	boolean
Confluence	Keep Newlines	Preserve newlines in extracted content.	No	`TRUE`	boolean
Confluence	Bins With Llm	Use LLM for processing binary files.	No	`FALSE`	boolean
Jira	Jql	JQL query to filter issues. If omitted, all accessible issues are indexed. Examples: `project=PROJ`, `parentEpic=EPIC-123`, `status=Open`.	No	None	JQL syntax
Jira	Fields To Extract	Additional fields to extract from issues.	No	None	array of field keys
Jira	Fields To Index	Additional fields to include in indexed content.	No	None	array of field keys
Jira	Include Attachments	Whether to include attachment content in indexing.	No	`FALSE`	boolean
Jira	Include Comments	Whether to fetch and index comments for each issue.	No	`FALSE`	boolean
Jira	Process Images	Whether to use a vision LLM to describe images referenced in issue descriptions and comments.	No	`FALSE`	boolean
Jira	Max Total Issues	Maximum number of issues to index.	No	1000	number
Jira	Skip Attachment Extensions	List of file extensions to skip when processing attachments (for example, `['.png', '.jpg']`).	No	None	array of extensions
Jira	Chunking Tool	Name of chunking tool for the base document.	No	`markdown`	`markdown` or empty

Search Index

Search indexed content using natural-language queries. This tool returns matching chunks from the selected index so you can inspect the raw search results directly. Purpose

Search indexed content with a direct semantic query.
Retrieve matching chunks from a selected index.
Test search quality before using more advanced stepback tools.

How to Execute

Open your toolkit.
Expand the Indexes section.
Open an existing index page.
In the left panel, select Select tool.
Choose Search Index.
Fill in the search parameters.
Select Search to run the tool.

Search Index form showing the query and parameter fields.

Common Settings & Parameters

UI label	Description	Required?	Default	Validation/Allowed
Query	Query text to search in index	Yes	None	None
Index Name	Target index collection name used for the search	No	`""`	max_length=7
Filter	Metadata filter for search results	No	`{}`	JSON format
Cut Off (0–1)	Minimum similarity score threshold	No	0.2	None
Search Top	Number of top results to return	No	10	None
Full Text Search	Full text search configuration	No	None	JSON with enabled, weight, fields, language
Extended Search	Additional chunk types to search	No	None	title, summary, propositions, keywords, documents
Reranker	Reranker configuration	No	`{}`	None
Reranking Config	Advanced reranking configuration	No	None	JSON with field weights and rules
Output Fields	Fields to include in the search response payload	No	None	page_content, score, metadata.*

Applicable to all toolkits that support indexing.

Stepback Search Index

Advanced search that rewrites or simplifies the query before searching and can use conversation context to improve retrieval. Purpose

Improve retrieval for complex or multi-part questions.
Use conversation context through the Messages parameter.
Return raw matching chunks after stepback query transformation.

How to Execute

Open your toolkit.
Expand the Indexes section.
Open an existing index page.
In the left panel, select Select tool.
Choose Stepback Search Index.
Fill in the search parameters.
Select Search to run the tool.

Stepback Search Index form showing the query and context parameters.

Common Settings & Parameters

UI label	Description	Required?	Default	Validation/Allowed
Query	Query text to search in index	Yes	None	None
Index Name	Target index collection name used for the search	No	`""`	max_length=7
Messages	Chat messages for stepback search context	No	`{}`	JSON format
Filter	Metadata filter for search results	No	`{}`	JSON format
Cut Off (0–1)	Minimum similarity score threshold	No	0.2	None
Search Top	Number of top results to return	No	10	None
Full Text Search	Full text search configuration	No	None	JSON with enabled, weight, fields, language
Extended Search	Additional chunk types to search	No	None	title, summary, propositions, keywords, documents
Reranker	Reranker configuration	No	`{}`	None
Reranking Config	Advanced reranking configuration	No	None	JSON with field weights and rules

Applicable to all toolkits that support indexing.

Output is a list of matching documents and chunks. If you prefer a generated answer, use Stepback Summary Index.

Stepback Summary Index

Contextual search plus an AI-generated answer, with optional citations. This tool combines stepback retrieval with answer generation for users who want a direct response instead of raw chunks. Purpose

Search indexed content with stepback retrieval.
Generate a concise answer from the retrieved content.
Support question-answering workflows where users want a direct response.

How to Execute

Open your toolkit.
Expand the Indexes section.
Open an existing index page.
In the left panel, select Select tool.
Choose Stepback Summary Index.
Fill in the search parameters.
Select Search to run the tool.

Stepback Summary Index form showing the query and answer-generation parameters.

Common Settings & Parameters

UI label	Description	Required?	Default	Validation/Allowed
Query	Query text to search in index	Yes	None	None
Index Name	Target index collection name used for the search	No	`""`	max_length=7
Messages	Chat messages for stepback search context	No	`{}`	JSON format
Filter	Metadata filter for search results	No	`{}`	JSON format
Cut Off (0–1)	Minimum similarity score threshold	No	0.2	None
Search Top	Number of top results to return	No	10	None
Full Text Search	Full text search configuration	No	None	JSON with enabled, weight, fields, language
Extended Search	Additional chunk types to search	No	None	title, summary, propositions, keywords, documents
Reranker	Reranker configuration	No	`{}`	None
Reranking Config	Advanced reranking configuration	No	None	JSON with field weights and rules

Applicable to all toolkits that support indexing.

If the answer lacks sources, increase Search Top or lower Cut Off slightly to include more candidate passages.

Remove Index

Delete an existing collection when it is no longer needed. Use this tool to remove outdated or temporary indexes from the toolkit. Purpose

Remove an index that is no longer needed.
Clean up outdated, duplicate, or test indexes.

How to Execute

Open your toolkit.
Expand the Indexes section.
Open the index page you want to remove.
Select Delete.
Confirm the deletion.

Delete Index confirmation dialog for removing an existing index.

Warning: Removing an index deletes its data from vector storage for this toolkit. Make sure you are targeting the correct collection.

List Indexes

List all available indexes for the current toolkit. Use this tool to confirm which indexes exist before searching, reindexing, or deleting. Purpose

View the indexes available for the current toolkit.
Confirm index names before running search or delete actions.
Check whether an expected index has been created.

How to Execute

Open your toolkit.
Expand the Indexes section.
Review the index list in the left panel.

List Indexes panel showing all available indexes for the current toolkit.

Search Tools Comparison

The following table compares the three search tools available in the Indexes section.

Aspect	Search Index	Stepback Search Index	Stepback Summary Index
Output	Raw results (document chunks with scores)	Raw results (document chunks with scores)	Short, AI-generated answer with optional citations
Uses conversation context	No	Yes (for query transform)	Yes (for search and answer generation)
Best for	Direct semantic search and quick result inspection	Exploring documents and improving retrieval for complex questions	Direct answers for end users
Cost/latency	Lowest	Lower	Higher

Tips & Troubleshooting

If search returns few or no results, try lowering Cut Off (for example, from 0.5 to 0.35) or increasing Search Top.
Use Index Name consistently (for example, prod, test) so you can target the right dataset later.
For better question answering, consider proposal-style chunking when indexing, because it turns on titles, summaries, and propositions.
For known limitations and workarounds, see Release Notes 1.7.0.

Guides & References

Indexing Overview
Create & Use Indexes
GitHub, Confluence, Jira, SharePoint — per-toolkit examples

Chat & Conversations

Indexing

Agents, Pipelines & Skills

Credentials & Toolkits

Entity Management

Prerequisites

Index Data

Common Settings & Parameters

Toolkit-Specific Settings & Parameters

Search Index

Common Settings & Parameters

Stepback Search Index

Common Settings & Parameters

Stepback Summary Index

Common Settings & Parameters

Remove Index

List Indexes

Search Tools Comparison

Tips & Troubleshooting

​Prerequisites

​Index Data

​Common Settings & Parameters

​Toolkit-Specific Settings & Parameters

​Search Index

​Common Settings & Parameters

​Stepback Search Index

​Common Settings & Parameters

​Stepback Summary Index

​Common Settings & Parameters

​Remove Index

​List Indexes

​Search Tools Comparison

​Tips & Troubleshooting

Prerequisites

Index Data

Common Settings & Parameters

Toolkit-Specific Settings & Parameters

Search Index

Common Settings & Parameters

Stepback Search Index

Common Settings & Parameters

Stepback Summary Index

Common Settings & Parameters

Remove Index

List Indexes

Search Tools Comparison

Tips & Troubleshooting